Skift Data + AI Summit: Five Tensions in the Room

Why It Matters

By spotlighting the practical obstacles to AI adoption, the summit equips travel firms with the insight needed to turn pilot projects into revenue‑generating operations, a critical step for staying competitive in a data‑driven market.

Key Takeaways

• Travel firms face a widening say‑do gap in AI implementation
• Expedia evaluates 1,500 AI agents to decide deployment priorities
• Only 6% of travelers trust AI to act on their behalf
• Marriott and Google Cloud stress production‑grade infrastructure for real‑world AI
• Skift summit frames AI decisions as tensions, not just topics

Pulse Analysis

The travel sector is at a crossroads where AI promises efficiency, personalization, and new revenue streams, yet many companies remain stuck in the planning phase. The Skift Data + AI Summit recognized that the real challenge lies not in technology selection but in bridging the "say‑do" gap—transforming glossy slide decks into production‑grade systems. By convening operators from hotels, online travel agencies, and cloud providers, the summit highlighted how infrastructure, data orchestration, and governance must evolve together to support scalable AI.

Central to the conference were five tensions that cut across every session. The first tension pits rapid pilot rollouts against the need for trustworthy, compliant models; Booking.com illustrated this by noting that merely 6% of travelers are comfortable letting AI make decisions for them. The second tension examines whether firms should build proprietary AI stacks or buy existing solutions, a debate echoed in Hilton’s warning that today’s stack may not scale tomorrow’s ambitions. Finally, ownership and scaling decisions—exemplified by Expedia’s systematic review of 1,500 AI agents—force leaders to prioritize resources and define clear accountability, preventing costly duplication and stagnation.

For investors and executives, the summit’s insights translate into actionable criteria for evaluating AI investments. Companies that demonstrate production‑grade infrastructure, measurable impact on the P&L, and a clear governance model are more likely to attract capital and achieve sustainable growth. As travel AI moves from experimental to operational, firms that resolve these tensions early will capture market share, improve customer experience, and set new industry standards. The Skift event thus serves as a practical roadmap, turning abstract AI hype into concrete, profit‑driving strategies.